Empirical Inference

16 papers14 directions
Empirical Inference is a research venue that has contributed to the machine learning and artificial intelligence literature between 2013 and 2024. With a corpus of 23 papers, the venue focuses on the theoretical and mathematical underpinnings of learning algorithms. Its publications frequently address core problems in statistical learning theory, including the development of PAC-Bayesian theory, the statistical analysis of Support Vector Machines (SVMs), and the direct approximation of divergences between probability distributions. The venue is characterized by its emphasis on optimization and learnability. Key research directions represented in its output include continuous convex relaxations for discrete submodular optimization and the study of sequential Rademacher complexity to determine online learnability. Additionally, the venue features work on online-to-batch conversion methods for statistical generalization and efficient transductive online learning using randomized rounding. The research often explores the formal properties of inference, as seen in studies regarding the median hypothesis and pre-image optimization in kernel methods. Empirical Inference serves as a platform for established researchers in the fields of statistical inference and computational learning. Prolific contributors include Bernhard Schölkopf, Masashi Sugiyama, Nicolò Cesa-Bianchi, and Ingo Steinwart, alongside other scholars such as Ran Gilad-Bachrach, Ohad Shamir, and Christopher J. C. Burges. Their collective work at this venue reflects a sustained interest in the formal properties of learning, ranging from historical perspectives on statistical foundations to contemporary developments in online and batch learning paradigms.

Papers by Year

2024 1
2013 16

Top Research Directions

Research directions most represented by papers from this venue.

Sequential Rademacher Complexity for Online Learning 1
Functional Gradient Descent for Ensemble Learning 1
Generalization Guarantees via Online to Batch Conversion 1
Incorporating Structural Dependencies into Sparse Regularization 1
Implicit Regularization via Early Stopping 1
Unified Mathematical Frameworks for Optimization Algorithms 1
Continuous Relaxation of Submodular Functions 1
Moving Beyond Binary Classification Frameworks 1
The Theoretical Gap in Complex Iterative Algorithms 1
Robust Boosting in the Presence of Label Noise 1
Non-Vacuous Generalization Bounds for Deep Learning 1
The Mechanics of Benign Overfitting in High Dimensions 1
Scaling Kernel Methods Beyond Quadratic Complexity 1
The Empirical Validation Gap 1

Top Authors

AuthorPapers
Bernhard Schölkopf2
David McAllester1
Dominik Janzing1
Gunnar Ratsch1
Ingo Steinwart1
Jonas Peters1
Joris M. Mooij1
Kun Zhang1
Leon Bottou1
Lorenzo Rosasco1
Marius Kloft1
Masashi Sugiyama1
Massimiliano Pontil1
Nicolò Cesa-Bianchi1
Ohad Shamir1
Ran Gilad-Bachrach1
Robert E. Schapire1
Samy Bengio1
Silvia Villa1
Tomaso Poggio1

Impactful Papers

Papers from this venue that shaped research clusters.

Influential

Papers

TitleYear
The median hypothesis2024
Direct Approximation of Divergences Between Probability Distributions2013
Efficient Learning of Sparse Ranking Functions2013
Efficient Transductive Online Learning via Randomized Rounding2013
Empirical Inference - Festschrift in Honor of Vladimir N. Vapnik2013
Explaining AdaBoost2013
In Hindsight: Doklady Akademii Nauk SSSR, 181(4), 19682013
Kernels, Pre-images and Optimization2013
Multi-task Learning for Computational Biology: Overview and Outlook2013
On Learnability, Complexity and Stability2013
On Sparsity Inducing Regularization Methods for Machine Learning2013
On the Relations and Differences Between Popper Dimension, Exclusion Dimension and VC-Dimension2013
PAC-Bayesian Theory2013
Semi-supervised Learning in Causal and Anticausal Settings2013
Some Remarks on the Statistical Analysis of SVMs and Related Methods2013
Statistical Learning Theory in Practice2013